Technology and Methodology

Unlike other social media listening services, Fizziology uses a mix of technology and trained statistics experts to collect, measure and score social media buzz. We've built a database that accurately tracks thousands of search queries around entertainment, sports, politics and brands and serves them up for our team of social media analysts to read, score and report on.

We pull directly from the source

We pull feeds from Twitter’s full Firehose of tweets and the APIs of Facebook and blog search engines. This allows us access to a higher volume of feeds on a more regular basis. There are free public search tools many of our competitors use, but they are not as stable or accurate.

We search for all relevant feeds – and only relevant feeds.

To create search terms for each property, we triangulate keywords -- meaning that we find and account for nicknames, contextual cues, abbreviations and common misspellings. For example, the movie "Mission: Impossible 4” chatter was searched for by also monitoring "MI4," "New mission impossible" and "Ghost Protocol." There are nuances between each search source and we have to develop the terms accordingly.

Using humans to research, create and evolve these terms has allowed us to accurately search for feeds for properties like NY Giants, Bridesmaids and Dove products.

For example, just using the search term "Rio" returns a high percentage of irrelevant results and would artificially inflate volume:

On the other hand, the search term Rio Movie means you'd miss a lot of conversation.

We may be capturing this:

But we're missing this:

We read, score and analyze posts.

Our database pulls in total volume and feeds from Twitter’s Firehose, real-time blog search engine Ice Rocket and Facebook’s API. Our social media analysts then read and score feeds to a 95 percent accuracy level. They score for sentiment (positive, negative, neutral or mixed) and can score for a relevant set of attributes, like "intent to see" for movies, "sexy" or "smart" for actors or "sporty" or "practical" for a car brand.

Real people read each feed so we can interpret misspellings and references, but also accurately score sentiment even with snark, sarcasm and slang.

We flag trends and provide actionable insights.

As analysts are scoring, they are also looking for trends in the conversation to report. What's driving positives? What's driving negatives? Is there a key line being repeated from a movie trailer? Or a marketing element being discussed for a brand? What does the audience look like (older, younger, existing or new fans, what brands do they associate with)? We use this information, along with volume and sentiment of comparable movies, teams, people and brands to provide an analysis for each property.

We map social media conversation to other marketing activity.

We can map spikes and falls in volume and sentiment of social media buzz to a property or brand's marketing and PR calendar to attribute what's driving positive and negative conversation, and what's not driving enough conversation. This allows marketers to make smarter decisions and better investments.